BFD Asymmetric Timers

Load base.ipv4.and.ipv6.cfg

#IOS-XE
config replace flash:base.ipv4.and.ipv6.cfg

#IOS-XR
configure
load bootflash:base.ipv4.and.ipv6.cfg
commit replace

Configure OSPF bewteen R4 and R6 using BFD. Configure timers as follows:

• R4 is willing to transmit as frequently as 750 msec, but only willing to receive as frequently as 1200 msec

• R6 is willing to transmit as frequently as 500 msec, but only willing to receive as frequently as 700 msec.

Will the BFD session form? If so, what timer values will be used?

Answer

#R4
int gi2.46
 ip ospf 1 area 0
 ip ospf bfd
 bfd interval 750 min_rx 1200 multiplier 3

#R6
int gi2.46
 ip ospf 1 area 0
 ip ospf bfd
 bfd interval 500 min_rx 700 multiplier 3

The BFD session will form. R4 will send at 750msec, and R6 will send at 1200msec.

Explanation

The first parameter for the command bfd interval is the min_tx interval. The min_tx is compared against the received echo interval to determine how fast to transmit echos. The reason for this, is that it allows a router to not be overwhelmed by echos if it cannot process them in the data plane. You don’t want a router to be able to just send as fast as it wants. It could potentially crash or cause instability for a weak BFD partner. This is why the min_tx is compared to the received interval, and the slowest timer is used.

The echo timers can be asymmetrical. Each end can transmit at different rates. There are two completely separate echos. The multipler is used locally to determine how many missed echos should result in declaring the BFD session down.

We can see on R4 that it is sending at 750msec. We can see that it received R6’s min_rx interval in the control packet (last line). The average echo values show that the rate is around 750msec. This is a bit confusing, because this only seems to be R4’s echo from its own perspective. It is sending them paced at 750msec, and receiving them back at this same pace.

On R6, we see that the echos are paced closer to 1200msec: